The present invention relates to data mining, and more specifically to utilizing online content to suggest item attribute importance to improve site level search.
Searching via natural language is a very desirable, user friendly method of searching. Natural language searching is designed to work ideally with data in a non-structured format. Unfortunately, most enterprise's systems of records have data in a structured format. A structured format data record from a product data catalog can be converted into a plurality of unstructured name/value pair attributes associated with the record. However, in some instances, too many attributes can actually obfuscate the search, resulting is less than optimal search results.